diffusion-reasoning/gdsd_code_llada
022
1{2 "activation_type": "silu",3 "alibi": false,4 "alibi_bias_max": 8.0,5 "architectures": [6 "LLaDAModelLM"7 ],8 "attention_dropout": 0.0,9 "attention_layer_norm": false,10 "attention_layer_norm_with_affine": true,11 "auto_map": {12 "AutoConfig": "configuration_llada.LLaDAConfig",13 "AutoModel": "modeling_llada.LLaDAModelLM",14 "AutoModelForCausalLM": "modeling_llada.LLaDAModelLM"15 },16 "bias_for_layer_norm": false,17 "block_group_size": 1,18 "block_type": "llama",19 "bos_token_id": 126080,20 "d_model": 4096,21 "dtype": "bfloat16",22 "embedding_dropout": 0.0,23 "embedding_size": 126464,24 "eos_token_id": 126081,25 "flash_attention": false,26 "include_bias": false,27 "include_qkv_bias": false,28 "init_cutoff_factor": null,29 "init_device": "meta",30 "init_fn": "mitchell",31 "init_std": 0.02,32 "input_emb_norm": false,33 "layer_norm_type": "rms",34 "layer_norm_with_affine": true,35 "mask_token_id": 126336,36 "max_sequence_length": 4096,37 "mlp_hidden_size": 12288,38 "mlp_ratio": 4,39 "model_type": "llada",40 "multi_query_attention": null,41 "n_heads": 32,42 "n_kv_heads": 32,43 "n_layers": 32,44 "pad_token_id": 126081,45 "precision": "amp_bf16",46 "residual_dropout": 0.0,47 "rms_norm_eps": 1e-05,48 "rope": true,49 "rope_full_precision": true,50 "rope_theta": 500000.0,51 "scale_logits": false,52 "transformers_version": "4.57.1",53 "use_cache": false,54 "vocab_size": 126464,55 "weight_tying": false56}57 